Menu Content/Inhalt
Home arrow Technology Park arrow IP Cameras arrow Digital Motion Tracking (ePTZ)

Digital Motion Tracking (ePTZ) Print
July 2012

This article is in the context of video surveillance pertaining to the tracking of a moving object within a predefined area beyond the traditional capabilities of a fixed head camera.  This introduction will immediately prompt us to think of a moving head camera called PTZ (pan-tilt-zoom) for the job.  Time has changed and technologies have improved.  Year 2012 is the year of Digital Video Surveillance coming of age.

The PTZ arrangement needs a person in the control room to watch the area via a locally installed camera in real time.  The person will need to control the camera head to follow the direction of the object being tracked.  If we do not have this 24x7 manning facility and depend on the preset PTZ tours of the camera, we cannot expect the criminal (assuming he is the object for tracking) will move in synchronisation with the camera head.  In brief, PTZ does not do the job well.

Two technologies have reached the stage of maturity as far as this application is concerned.  The first is the camera viewing and recording intensity.  The second is artificial intelligence.

Mainstream cameras with embedded Internet Protocols hit 1 megapixel back in 2008.  They hit 4 megapixels in 2012 and their area coverage would be 4 times as in 4 x 1MP.  They will not be 2x2 MP because video display is not in a square shape.  Let us assume that the 4MP camera provides a long rectangular 180 degree panoramic view.  See the picture as an illustration.

Artificial intelligence comes into play by breaking a single 4MP frame into 4 smaller frames with 1MP each.  Let us call the frames A, B, C and D.  When a criminal walks from A to B, C, and D, he could be seen in 4 individual frames or in one single large panoramic view. The edges of a panoramic view are distorted for the same reason that the sides of the Earth on a 2D map are not as clear as the central part of the map.  When the criminal is in location A or D, it may be hard to see him clearly.  This does not apply to a small frame such as A or D because distortions are minimal.  A small frame still has 1MP and can be enlarged to fill the screen.  Correction of minor distortions is available through digital means too.

Do we manually switch frames to keep track of the criminal?  No, we can apply automatic motion detection techniques.  We define what constitutes motion for each small frame and set the priority of display in case there is more than 1 frame with motion detected.  The correct frame will be displayed at the right time in sequence.  This allows recording to take place in the background without any manual control.

As of July 2012, Compucon is able to supply such level of technologies for video surveillance purpose.  We will set each small frame to 1280x760 resolutions, 6 frames per second, 750kbps of bandwidth, and use minimal hard disk space for recording.  We can do this for indoor conditions now and will cover outdoor soon.